Hacker News new | past | comments | ask | show | jobs | submit login
The Early History of Usenet, Part I: The Technological Setting (columbia.edu)
123 points by longdefeat 17 days ago | hide | past | web | favorite | 39 comments

Back in the early 90s, when a fast modem was still excruciatingly slow, a webpage could (and did!) take tens of minutes to load. I tried downloading a six megabyte file once which used up about 4 hours of my monthly quota, only for the file transfer to be aborted right near the end. I nearly cried.

Most of us then used UUCP to send and receive our email, usenet, etc from the Net. A typical call lasted 15 to 20 minutes sometimes.

Only the rich power-users could afford to log in more than once a day, it was too expensive otherwise. And the service providers were charging an arm and a leg per hour, with a monthly contract total of say 30 hours.

Around that time, one of the local members of the user-group posted to say that he had sent an email across the Net to the US, and had received a reply within 10 minutes! We were astounded at the speed of that turnaround.

Also about then, somebody told us that they had heard that the total number of computers on the Net had reached a whole million. Again, we were astounded. We literally couldn't envisage that within a decade, that number would reach billions.

And don't get me started on how twitchy I got when the 'always-on' broadband era arrived. I kept getting thoughts of thousands and thousands of dollars of provider charges racking up every month. <grin>

I was on the net in the early 90's and before, and maybe I'm wearing my rose-colored glasses today, but I don't recall any web pages taking anywhere close to ten minutes to load. Maybe a minute at worst. I don't think I'd have had the patience to browse the web otherwise.

Though computers were less powerful and network speeds were comparably slow, web pages were mostly really lightweight back then, not the bloated monstrosities we have today (though even then there were some atrocious ad-infested websites, which were slow.. but not 10 minutes per page slow, in my experience).

Also, I seem to recall that downloading my first mp3, which was about 3 MB in size took around 30 minutes back in the day.

I don't doubt that downloading that 4 MB file and viewing that webpage was as slow for you as you reported, but I do wonder if maybe you were using a really old modem for they day. I think by the time I first started accessing the internet I must have had a 9600 baud modem, and not many years later switched to 56k ISDN. How about you?

PS: I just remembered that pretty early on in the web's history, I read about some study which found that if a website responded in more than 200 milliseconds, most people considered that slow, and I remember thinking that that observation fit pretty well with my experience. So I think pretty much from the beginning websites must have overall been responding in well under a second as a rule, and even back in the day taking a full minute (nevermind 10 minutes) would have been considered unacceptably slow.

Yeah...back in the 90s, a webpage was a text file, usually a couple of Kb tops, maybe a small gif or three. No pulling in Mb of frameworks, no client side executable content, just plain content. If it took 10m to load at 9600, you were doing it wrong (or were targeting the folks with DDS/T1/T3 connectivity). If you were serving out a big file, you often did it over FTP.

> I tried downloading a six megabyte file once which used up about 4 hours of my monthly quota, only for the file transfer to be aborted right near the end. I nearly cried.

The ability to resume an interrupted transfer was a huge benefit in the ZMODEM protocol, as compared to its predecessors XMODEM and YMODEM.


There were lots of innovations in this area, many of them having to do with how unreliable modem connections could be (both in terms of possibly un-hardware-corrected line noise, and in terms of the risk of having your connection hang up in the middle).


I sure felt safer when I was downloading a big file with ZMODEM rather than XMODEM. :-)

This nostalgia-laced post deserves an upvote just for the closing <grin>. I thought that was lost to the ages...

I miss my 9600 Baud Hayes external modem...

Kids these days won't understand the feeling that you got when the modem connection noise sounded right, and you knew you were going to get a good speed, and when it sounded off – even though you eventually connected at a certain speed, you didn't trust it...

I miss my 9600 Baud Hayes external modem...

9600! Looxery, Lad! For many years 300 baud was the maximum I had. Then 1200, 2400, and at long last 9600! And those damned things were hundreds of dollars.

deserves an upvote just for the closing <grin>

One of the local UUCP group mentioned above habitually used <smirk>.


I remember when modems cost about a dollar per baud: 300 for $300, 1200 for $1200.

I remember when every incremental advance in modem speeds and standards was a big thing. Companies like Sparco[0] were major movers.

At home, moving up from a 300/1200 external modem to a 9600bps internal (backplane card) was huge.

I was first on Usenet in 1986. My employer didn't have a full-time connection then, just a modem that would autoconnect every few hours to upload and download email/Usenet/etc. This was before domain addressing, where you would have to give a full navigation path from a well-known server like decwrl or uunet. I was literally the first person in my company to put my long form email address on my business cards (before @(domain) addressing).

[0] https://sites.google.com/site/unistarsparcocomputers/

The first FLOSS I ever contributed to was 2.11B News, the standard Usenet software of the day, back in 1985 or so. I ran a Eunice system (a kludge that ran Unix on top of VMS) for a startup, and wanted to get us on Usenet; earlier I had a job at a Navy lab that was on the ARPAnet and felt cut off. Larry Wall (himself) had done some Eunice patches for an earlier version, and I ported them to work on the new system. The only tricky part was faking links, since Eunice didn't support them. 300 baud modems and magnetic tapes by mail were the only way to move your data around.

To this day, the Configure script for Perl still outputs "Congratulations. You aren't running Eunice."


Wow, I had somehow forgotten about those chatty diagnostics about the Perl build system's reasoning about your local OS (maybe due to not having had to compile Perl recently?). They're awesome.

Classic Perl Journal, September, 1998:


I was amazed by Usenet when I first encountered it. Communicating with people around the globe really seemed like the future had arrived. The company I worked for would connect to the University of Rochester via a Telebit Trailblazer modem and from there U of R would connect to the Internet. I recall our sysadmin complaining the daily bundle was nearing 80 Meg and was straining disk space on our Sun server.

Absolutely riveting article. But I must note that the page is quite obnoxious to read on a mobile device. I'd prefer the content alone and none of the decorations around it..

E: uBlock Origin rule to block the footer on e.g. Firefox Android:


Usenet is probably the most decentralized medium out there. If I am not mistaken, this how the protocol worked ... data was posted to one server, then every other server replicated the data for each client to consume. Typically the server had like a years retention for binaries and an archive for text-based stuff.

Usenet is sorta like mastodon and Reddit combined.

Originally the UUCP protocol was used, and most Usenet sites had no network connection other than a dialup modem (other than perhaps a local Ethernet, if that). Retention time was only a few days: available disk was very limited. Later, the NNTP protocol was developed to transfer Usenet over TCP/IP, which is what you're thinking of.

The UUCP protocol was still used to enable the replication he described. Also, it was not black and white, UUCP or NNTP. For example, for a while I ran a local NNTP server and received news through UUCP.

Bram Cohen explicitly cited Usenet's distributed model as the inspiration for Bittorrent's design.

I made my first internet (or was it arpanet?) purchase in about 1986, using Usenet. Usenet had for-sale listings, subdivided by region; I was working at Boeing Computer Services (part of Boeing, but a part that no longer exists), and had Usenet access. My wife said she wanted a dishwasher. I told her she had married one, but that didn't seem to work.

One day I saw an ad for a portable dishwasher on the Usenet for-sale in my area. Contacted the guy (email? phone? I can't remember). We connected, and I bought it. I think both our wives were astonished at what computers could do.

Took the dishwasher with us to Colombia a couple years later, and sold it a few years later for what I had paid + shipping. Now say that about your modern internet purchases!

This brings back memories. I ran a UUCP node when I was in high school, back in the early 90's. I had a feed from another local, who connected to a system somewhere on the west coast. This gave me internet email and about a dozen newsgroups...

I loved USENET. Still do for the few comp.sys groups I visit and that are still active.

I loved how efficient it was. Unfortunately, that was also the downfall as it allowed people to spam efficiently.

I'm still bummed that everything went to the web instead of trying to fix the moderation problem on the Usenet. I have never seen an online discussion forum as fast and featureful as my old Usenet client.

Reddit, despite its own flaws, is the closest thing to Usenet in terms of a text-oriented display with keyboard-heavy navigation (with Reddit Enhancement Suite).

Reddit is definitely closest to the old Usenet experience, but its search engine and filters leave something to be desired.

I prefer uncensored.citadel.org. Yes, no https support, but you can login over SSH with ssh bbs@uncensored.citadel.org and use it as a modern BBS. It works great.

I'm not in love with the interface to be honest. It's great if you want to consume absolutely everything, but if you want to pick and choose in a topic thread there isn't a summary view as far as I can tell. It's kind of like reading specific #hashtags on Twitter.

warning for other users: i randomly clicked on the israel subforum and politics one and... yikes. chan-level bigotry

I love finding these retro-net communities. Another good one is the bboard at sdf.org.

killfiles, and you are done. Even better, slrn has a TUI assistant.

My ~/.jnewsrc













> killfiles, and you are done.

no, that didn't work. and that's a lot of the reason that usenet died.

no matter how many of the regulars in any given group might killfile the local trolls, there was always at least a few people that did almost nothing but spar with them, filling the group with information-free dross that nobody sane was interested in. and there was nothing stopping the trolls from mutating their email addresses frequently, in an attempt to evade killfiles.

hacker news is more-or-less drama-free because the mods have the power to remove disruptive users. if you'd try to apply a killfile-like solution here, HN would be long dead by now.

And real, no hassle, threaded discussion!

I have similar thoughts. Perhaps solving that problem would still count.


More realistically, for all the places where a web-of-trust model fails moderation seems like the perfect fit for it. I see no reason we can't have multiple teams of moderators whitelisting/blacklisting different content in the same namespace, with their changes applied only to users who opt-in.

Federated accounts between servers might be a nice overlay feature set. Servers can just reject mail from any unauthenticated user, and sync user databases periodically.

Federated moderation enables extremists to blacklist people who endorse something they don't personally like and limit their ability to use other sites.

This is a plot point in Neal Stephenson's "Fall".

The articles get better and better, can't wait for part IV.

"PC Pursuit" was a flat rate service offered by GTE Telenet that let you side-step long distance fees by dialing into Telenet locally and dialing out on banks of modems in other cities, to access BBSs, UUCP servers, etc.


Remotely accessible shared modems had some amusing security flaws: "PC Roulette" was a fun game you could play by connecting to a modem in some city and typing "A/" to redial the last number it called, to discover all kinds of weird random BBSs, Unix boxes, FidoNet nodes, and other services.


>In the late 1980s, Telenet offered a service called PC Pursuit. For a flat monthly fee, customers could dial into the Telenet network in one city, then dial out on the modems in another city to access bulletin board systems and other services. PC Pursuit was popular among computer hobbyists because it sidestepped long-distance charges. In this sense, PC Pursuit was similar to the Internet.


>PC-Pursuit (Telnet) flat rate off-peak modem calls


>"Doing PCP all night can be legal. ;-)"


>[Moderator's Note: There is an interesting history behind the whole thing. Prior to about 1984 when PC Pursuit began operation, Telenet had their data network going, which dates from sometime in the 1970's. Like the phone network, it was busy all day and almost deserted all night. Telenet started PC Pursuit as a way to make use of all the facilities sitting idle all night long. I was one of the first half-dozen or so users to sign up for PC Pursuit when it started operation back then. They used a clumsy, rather tedious call-back system where you dialed in, entered your (authorized) call-back number, disconnected and waited for their return call to put you on the network. There were about five cities we could call in the beginning, at 300/1200 baud only. PC Pursuit was greatly improved upon as the years went by. For many years they even offered unlimited access between 6 PM and 7 AM for $25 per month. It was such a good deal they eventually had to put limits on the amount of time people could use the service each month without extra payment. I would not be surprised if they are now swamped beyond their capacity to handle the traffic. PAT]


>The PASSWORDS are varied. They are initially registered to the user in the format of XXXXyyyy where XXXX are four letters and yyyy are four digits...

>NOW, your saving grace from people changing their passwords is that Telenet charges $5.00 for each password change... So this helps...

What a great business model: charge customers $5 to change their password! Why didn't I think of that???


Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact